Determining Forest Structural Attributes Using an Inverted Geometrical-Optical Model

The Montreal Process indicators are intended to provide a common framework for assessing and reviewing progress toward sustainable forest management. There is a need for techniques using available remotely sensed data to evaluate these indicators. This project investigated the potential of a combined Geometrical-Optical/spectral mixture analysis model for forest assessment mapping of the Montreal Process age class and successional stage indicators at a regional scale using Landsat Thematic Mapper data. The project location is an area of eucalyptus forest in Emu Creek State Forest, Southeast Queensland. A quantitative model relating the spectral reflectance of a forest to the illumination geometry, slope and aspect of the terrain surface and the size, shape and density of the trees has been developed. This combined model exploits both the spectral and spatial variance in a digital multispectral image to uncover biophysical information. In the case of the model developed, estimates derived were of crown-cover projection, tree density and canopy size. Inversion of this model necessitated the use of spectral mixture analysis to recover subpixel information on the fractional extent of ground scene elements. The spectral mixture analysis utilised information on the spectral characteristics of individual ground scene elements (such as sunlit canopy, shaded canopy, sunlit background and shaded background) and assumptions about their spectral and spatial interactions. Explicit formulations of both Geometrical-Optical and spectral unmixing models allowed the effects of input parameter perturbations to be quantitatively determined in a sensitivity analysis. By applying knowledge of the main controlling factors, the sensitivity analysis allowed improved allocation of resources to maximise the predictive accuracy of the model. It was found that modelled estimates of crown cover projection, canopy size and tree densities had significant agreement with field and air photo interpreted estimates, however, the accuracy of the successional stage classification was limited. The results obtained highlight the potential for future integration of high and moderate spatial resolution imaging sensors for monitoring forest structure and condition.